Professional Documents
Culture Documents
A R T I C L E I N F O A B S T R A C T
Keywords: Part of the Harbor Bay Business Park, Alameda in California, showed evidence of liquefaction during the 1989
Earthquake-induced liquefaction Loma Prieta earthquake, but this was accompanied by only small settlements and no significant damage to two
Critical state soil mechanics recently constructed office buildings. Digital CPT records are available from a reasonably comprehensive post-
State inversion
earthquake investigation. These, in combination with the large body of test data associated with the Park’s
Loma Prieta earthquake
development, have been used to assess the soil behavior at this site further. Using analysis based on critical state
soil mechanics, it appears likely that the presence of underlying Bay Mud was an important factor, with the
limited undrained strength of that stratum limiting the transmitted shear stresses to the overlying and liquefiable
sand fill but without preventing its liquefaction. This inferred behavior is consistent with a system (i.e., favoring
interaction between layers) response, which was also observed during the 2010–2011 Canterbury earthquake
sequence.
1. Introduction fill and the underlying Bay Mud, with conventional reliance on surficial
manifestations of liquefaction (there were no strong motion or
This case history is within the Harbor Bay Isle project, which consists piezometer records at this site).
of 940 acres of hydraulic fill dredged from San Francisco Bay and The ITP area is enclosed by a South Loop Road and the Harbor Bay
pumped into the site after constructing perimeter dikes, starting late in Parkway, with an aerial view and site plan shown in Fig. 1. Fig. 2 il
1966. Harbor Bay Isle is adjacent to the south, west, and north sides of lustrates one of the two buildings constructed in the ITP before the 1989
Bay Farm Island, a district of Alameda city, California. Both Harbor Bay Loma Prieta earthquake, corresponding to a standard low-rise com
Isle and Bay Farm Island are separated from the rest of the city on mercial development. The extent of liquefaction shown in Fig. 1 is from
Alameda Island by the San Leandro channel. About 600 acres were an inspection of the time (Hallenbeck & Associates [3]) and the lique
zoned as residential and the remainder as commercial; the latter is now faction details are discussed in the manuscript; the buildings shown on
known as the Harbor Bay Business Park (HBBP). The HBBP borders on the photograph to the west and northwest of the ITP had not been
the Oakland International Airport and the San Francisco Bay. constructed in 1989, and liquefaction there was a greenfield condition.
There were instances of liquefaction and non-liquefaction within
Harbor Bay Isle during the magnitude 6.9, 1989 Loma Prieta earthquake 2. Methodology
that affected California. Liquefaction in the Harbor Bay Business Park
has been referred to as Site 4 in Mitchell et al. [1] and as Alameda Bay The HBBP case history is developed using critical state soil me
Farm in Moss et al. [2]. However, these prior investigations used limited chanics, and it is helpful to summarize this framework as it differs from
cone penetration test (CPT) data and also, as conventional at the time, the current predominant practice. For sands, the current practice largely
regarded the hydraulic fill as the only soil of interest. Here we consider uses the so-called NCEER method for assessing liquefaction triggering (e.
in more detail a 0.12 km2 part of the HBBP, which was originally called g., Youd et al. [4]) where fines content (measured or typically inferred as
Plaza 6 and then renamed the International Teleport Plaza (ITP). Our in the case of the Cone Penetration Test – CPT) is used in addition to
approach follows a critical state framework for both the dredged sand penetration resistance. If sufficiently plastic, fine-grained soils are
* Corresponding author.
E-mail address: jorge.macedo@ce.gatech.edu (J. Macedo).
https://doi.org/10.1016/j.soildyn.2022.107280
Received 22 August 2021; Received in revised form 1 February 2022; Accepted 25 March 2022
Available online 27 April 2022
0267-7261/© 2022 Elsevier Ltd. All rights reserved.
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Falls [5]. The engineering of this dam was based on the recognition that,
as soil comprises particles, the distortion (“shearing”) of soil generally
involves a change in the void ratio (e) as the particles move past each
other. If the soil is loose, the particles will tend to move to a denser state,
while the opposite will occur if the soil is dense. Two particular un
derstandings then developed: i) void ratio evolved to a unique
critical-state value with distortion regardless of its initial value (Casa
grande [6,7]), with this large-distortion value defining the critical state
line (CSL) and being the key to understanding soil behavior as any soil
that was looser would be contractive and so liquefiable if saturated
(hence the name ‘critical void ratio’); and, ii) soil strength developed
from the combination of ‘friction’ (i.e., the critical state friction) and
‘interlocking’ (i.e., dilatancy caused by distortion) which was formalized
by Refs. [8,9]. The missing idea at that time was how to link (i) and (ii);
this link was provided some forty years later and was that the limiting
dilatancy depended on the difference between the soil’s current void
ratio and its critical void ratio at the current mean effective stress, which
was defined as the state parameter (ψ ) by Been and Jefferies [10]. These
three ideas are key for a modern understanding of soil behavior and
apply regardless of a soil’s gradation, fines content, etc. (Jefferies [11]).
In particular, this framework was found necessary to understand recent
liquefaction failures of three large dams: Fundão [12]; Cadia [13]; and,
Brumadinho [14,15]. The critical state approach has also been found
helpful in unifying cyclic liquefaction data (e.g., Refs. [16–20]).
In applying the critical state approach to the HBBP case history, the
fines content’s notion is replaced by the soil’s stress ratio at critical state
- also known as soil’s intrinsic friction-i.e., Mtc = q/p, with q the
deviatoric stress and p the mean effective stress at critical state), its
compressibility (λ10 ), and additional mechanical properties (discussed
in section 4). Of note, under a critical state framework, a ‘sand like’ or
‘clay like’ separation is not required; different soils simply exhibit
different mechanical properties such as Mtc , λ10 . Correspondingly, one of
Fig. 1. International Telegraph Plaza layout. Representative subsurface infor
mation is shown in Fig. 3, a) Aerial view of the HBBP showing the ITP location the key aspects of the case history discussed in this study becomes
where liquefaction occurred during the 1989 Loma Prieta earthquake, b) Plan assessing the properties of the soils involved. A further difference to the
of development from ~1985 showing site investigations and pro NCEER methodology is that in a critical state approach, CPT data is not
posed buildings. reduced to a reference stress level, but instead, the data is used as
dimensionless stress-ratios in a formal inversion following the laws of
mechanics (the inversion process is discussed in section 5). Thus, the
final output is the site response being characterized in terms of the in-
situ state parameter, which is linked with mechanical soil properties.
These aspects are illustrated later in the manuscript.
3. Ground conditions
2
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Fig. 3. Stratigraphy at HBBP from a deep CPT record (data courtesy of Langan).
3
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
clay mixtures, the clay being plastic: Layer 4. There were a few water
contents of 50% and above in layers logged mainly as sandy, but these
are believed to be isolated pockets. The boring logs also indicate that the
soft, plastic clay “slurry” was also encountered as thin seams as well as
thicker layers.
The boundary between the fill (about 20 years old in 1989) and the
underlying natural soils is the top of Layer 5, the Holocene Bay Mud. The
HBM was encountered in most of the borings, except for those near
Harbor Bay Parkway. The transition elevation did not vary much,
ranging from about 26.2 to 27.1 m and averaging 26.7 m with a standard
deviation of 0.4 m. This stratum thickens, but not smoothly, from
northwest to southeast across the ITP.
The top of layer 6 is the transition to much older soils and soils found
elsewhere in the region. Layer 6 is a silty sand, called Merritt Sand by
other workers (e.g., USGS [23]). It was encountered in all CPTs. The top
of layer 7 is the start of the PBM sequence. There are multiple sub-units
within the underlying Pleistocene Bay Mud identified as Layers 7, 8, and
9. These layers comprise clay and silt strata to a considerable depth; the
full depth of this unit was not identified in the HBBP, but it has a
thickness of some 40 m at the Naval Air Station.
4. Soil properties
4.1. Sand fill (layers 1 and 2) Fig. 5. Critical state locus of the HBBP fill sand and the Isserk sands with
gradations of 5% and 10%.
The hydraulically placed sand fill has a median grain size range of
160 <D50 < 220 μm and a fines content of less than 20% (typically about encountered with other sands (Jefferies and Been [25]). This feature will
10%), illustrated in Fig. 4. A sample of this fill collected in 2018 was be used subsequently on identifying the in-situ state of the fill sands (see
used for triaxial testing to determine the mechanical properties, section 5).
following the protocols given in Jefferies & Been [24]. The CSL deter The corresponding strength data representative of the fill materials is
mined is shown in Fig. 5. shown in Fig. 6, with the strength properties Mtc (already defined) and
The triaxial testing on HBBP fill was supplemented using data on N, which is the volumetric coupling in a strength-dilatancy relationship
similar sands, in particular, to quantify the effect of changing fines (q/p)max = Mtc − (1 − N) Dmin , where Dmin is the maximum dilatancy.
content on the fill mechanical behavior. A large body of fill testing was The effect of soil state on the limiting (“peak”) dilation rate is shown in
carried out for oil exploration in the U.S and Canadian arctic offshore, Fig. 7 through the state-dilatancy parameter X, which represents the
deriving on data on the public domain (Jefferies and Been [25]). This scaling of Dmin and the soil state, using the state parameter (ψ ) as defined
data was searched, and Isserk sand, a predominantly quartzitic sand in Been and Jefferies [10]. Mtc , N, and X can be calculated from triaxial
with no plastic fines as the collected fill, was identified as gradationally tests with procedures detailed in Jefferies and Been [25], which we have
similar to the HBBP fills (see Fig. 4). Interestingly, the CSL of Isserk sand
with 5% fine contents (indicated as Isserk 220/5 in Fig. 5) is also
consistent with that of the collected fill (see Fig. 5), which also has
similar fine contents. The tests on Isserk sand with 5% and 10% fine
contents are used in this study to include gradational effects on assessing
the in-situ state of the fill sands. For instance, the CSLs of the two Isserk
sand gradations are shown in Fig. 5 to illustrate the effect of changing
fines; increasing fines increases the slope of the CSL (λ10 ), a behavior
Fig. 6. Effect of the maximum dilatancy (Dmin ) on the drained strength of the
Fig. 4. Gradation envelope of HBBP fill, recovered HBBP sand, Isserk sand, and sand fill. ηmax corresponds to the maximum η during each test, and Mtc is the
examples of fill with slurry inclusion. stress ratio at the critical state.
4
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Fig. 8. Insitu void ratio of Holocene Bay Mud. The CSL and ψ contours are
approximated to match the observations in Figs. 18 and 19 and are shown only
for illustrative purposes. The CSL is estimated by recovering a ψ = − 0.05
(Fig. 19) for an initial void ratio of 2.04 at a mean pressure of 100 kPa (Fig. 18).
Fig. 7. Control of maximum dilatancy (Dmin ) by state parameter (ψ ).
have found similar normally consolidated conditions in the HBM else
followed in this study. Sand properties are summarized in Table 1, with where (e.g., Refs. [26–28]). These conditions are reflected in the void
the meaning of the properties annotated on the respective figures. ratios of the in-situ HBM samples being predominantly looser (wetter)
than the estimated critical state locus (see below), as shown in Fig. 8.
The initial compression index (Cc ) was largely in the range 1.05 < Cc <
4.2. Compromised fill (Interbedded layers 3 and 4)
1.37 with a representative Cc ~ 1.10. However, most unusually, the
swelling (elastic) index was small with a consistent Cs /Cc ~0.1; this is
Layers 3 and 4 appear as intermingled soils within the extensive
about half the elastic void ratio recovery normally expected for soft clay.
Layer 2, with questionable lateral continuity. There are no triaxial tests
The strength tests were only performed on samples that did not
or similar on either 3 or 4 layers, so the properties must be estimated. In
display visible signs of disturbance. Strength was taken as the peak of the
the case of Unit 3, its compressibility is assessed (see section 5) from the
stress-strain curve or the stress at 10% strain, whichever occurred first.
CPT data based on its ‘soil behavior type’. Unit 3 appears to retain the
The data is shown in Fig. 9, normalized by the in-situ vertical effective
characteristics of the parent sand fill but with increased compressibility
stress (σ v0 ); the undrained strength ratio lies in the range 0.2 < su /σv0 <
′ ′
5
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
5.2. Comparison of CPT before and after the 1989 Loma Prieta
earthquake
6
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Fig. 11. Comparison of paired CPT resistance profiles before and after the Loma Prieta earthquake.
The last overlay of Fig. 11, the ITP-7 pair, gives an opposite view
with clear densification of the zone annotated as B.
It is standard to use post-earthquake measurements as indicative of
the prior conditions (e.g., Mitchell et al. [35]), and that is forced for
detailed analysis of this case history because only CPT campaigns later
than 1989 had the needed digital data acquisition. The comparison of
this digital data with the older paper records suggests that, on average,
this is acceptable but that the 1989 records will have greater dispersion
of the tip resistance within the loose sand fill. As far as can be judged,
because of the limited available resolution of the 1984 data, the HBM
shows a similar response in 1989 as 1984.
The results of the 1989 CPT soundings within the sand fill, layers 1
and 2 on Fig. 3, are shown in the Jefferies and Been [24] ‘soil behavior
type’ chart in Fig. 12. This chart also shows “screening” level contours of
ψ , which are useful for an initial appreciation. The upper 2 m or so,
Layer 1, is the ‘ubiquitous dense’ stratum across the site and is very
dilatant (ψ < − 0.2) and presents as clean sand. In parts, this layer is very
dense. It is above the water table, and thus its only role in the lique
faction is as an overlying stiff mass. The underlying sand fill, Layer 2,
plots as rather heterogeneous with soil type ranging from ‘sand, some
silt’ through to ‘silty sand’; exactly as expected from the gradations
determined on the fill samples. The data tends to follow a constant-state
Fig. 12. CPT data within the hydraulic fill as a soil behavior type.
contour (ψ ≈ − 0.1), an aspect of soil development/accretion that has
been found elsewhere (Reid and Jefferies [36]); in essence, the soil
naturally compensates for the changes in gradation from place to place
7
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
The south side of the ITP is underlain HBM, and the results of the
1989 CPT soundings within the Bay Mud are shown in Fig. 17. This
stratum plots as a brittle, sensitive silt, rather different from the
perception of a plastic normally consolidated soft clay. There are sandy
stringers within the HBM, which present as loose, but that is most likely
an artifact of having soft silt above and below (no ‘thin layer’ correction
has been used).
The NorSand implementation of critical state theory was calibrated
to the Bay Mud, the calibration being shown to monotonic simple shear
data reported by Rau and Sitar [27]) in Fig. 18. Test MONO-8 was
chosen as that was one of the loosest tested with a void ratio e = 2.04 at
an effective vertical stress of 100 kPa, a void ratio and stress combina
tion that is within the range of the in situ conditions at the site albeit at
the denser end of the range. The calibration adopted λ10 = Cc (not un
reasonable for a fine-grained material and Cc having been measured, as
Fig. 13. Example of computed CPT inversion considering various elastic ri discussed earlier), Cs /Cc = 0.1 (characteristic of measured values),
gidity (Ir ) and initial mean pressure (p) values. Mtc = 0.95 (measured, discussed earlier), N = 0 (a ‘not unreasonable’
8
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Fig. 17. CPT data within the Holocene Bay Mud as soil behavior type.
Fig. 16. Inverted ψ below water table within the fill at ITP from 1989
CPT soundings.
value for silts) and over-consolidation ratio OCR = 1 (as reported by Rau Fig. 18. Calibration of NorSand (red) to Bay Mud Test MONO- 8 (blue) in
and Sitar [27]). Other properties, and the state parameter, were then monotonic simple shear tests. The MONO tests come from Rau and Sitar [27].
iterated to get a representative fit of the measured stress-strain curves;
the achieved fit was for a lightly dilatant state ψ = − 0.05, a
Jefferies [31], with an updated version of the software for direct use
state-dilatancy coefficient X = 2, and a plastic hardening modulus H =
with u2 data; this update was validated in the Cadia silt (see Figure E4-5
6.
of Morgenstern et al. [13]). The computed inversion coefficients, using
The calibrated soil properties were then carried forward into the
the HBM properties as per Fig. 18, were k’ = 6.5, m’ = 2.6, and the
inversion of the CPT data to compute the in situ state. The methodology
results of processing the CPT data with these coefficients are shown in
uses the measured CPT data, which is normalized using the excess pore
Fig. 19.
pressure at the u2 location:
A first point to note in Fig. 19 is the state used to calibrate test DSS-
′
m ψ = − ln(Q / k )
′
[2] MONO8 (Fig. 18), which is well-aligned with the denser computed in-
situ state and consistent with that test being at the low end of the in-
where situ void ratios. The in-situ state range computed from the CPT is also
( ) consistent with the data in Fig. 8. Going further, the “classical geolog
Q = Q 1 − Bq + 1 [3] ical” idealization of ‘normal consolidation’ is for the NCL to parallel the
CSL in e − log(p’) space with an offset of about a factor of 2 measured on
and k’, m’ are the soil-property-dependent inversion coefficients. These the p-axis; if this is converted to void-ratio, ‘normal consolidation’
coefficients were computed using the cavity inversion in Shuttle and amounts to ψ ~ 0.8/(λ − κ), where λ is the CSL slope in natural log units
9
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Data for the older sediments come from more recent (2018) CPT
soundings in an adjacent area of the HBBP, where three soundings were
pushed to some 25 m depth. The results are shown as soil behavior type
in Fig. 21.
The Merritt Sand presents as dense sand, some silt grading to silty
sand in parts. The soundings in this unit showed negative excess pore
pressure (see Fig. 3), indicating at least only partial drainage; this sug
gests this unit is gradationally finer than indicated by its soil behavior
type. The density of these sands suggests a transgressive marine depo
sition with wave-induced densification; at a ‘screening level’, the Merritt
Sand exhibits − 0.10 > ψ > − 0.20, although this range reflects system
atic change across the site with CPT-2018-11A encountering looser
conditions (ψ k ~− 0.12) than CPT-2018-10A (ψ k ~− 0.17) throughout
the Merritt and with the third CPT-2018-09A lying in the middle.
As previously discussed, there are multiple sub-units within the un
derlying Pleistocene Bay Mud. Layer 7 presents as over-consolidated
and/or dense to the extent of being quite dilatant; the undrained
Fig. 19. Computed in-situ state of Bay Mud. strength ratios commonly exceed su /σv0 > 0.7 using Nkt = 16. Layer 8,
′
Fig. 20. Computed undrained triaxial compression of Bay Mud using Norsand Fig. 21. CPT data within the Pleistocene sediments, plotted by soil
calibrated properties. behavior type.
10
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
with a recording strong-motion station at the time was the Naval Air Center and surgical dynamics sites, and along the north side of Harbor
Station (NAS; VS30 ≈ 150 m/s), about 7 km distant, and the outcrop Bay Parkway north and west of the ITP. Sand boils were apparent around
motion at YBI correlates with that of the surface at the NAS with an the ITP buildings and in the undeveloped areas throughout the 0.12 km2
approximate time delay of 320 ms, illustrated in Fig. 22. Of note, the ITP site, see Fig. 1. There were several locations in the HBBP where thin
same stratigraphic sequence is found at HBBP and NAS; in addition, the cracks were apparent in the ground surface. Some of these cracks
VS30 are similar for the two sites, thus we will use it as the surface ground extended for substantial distances and were side by side and acted as
motion. The motion of the NAS site stays aligned with that of the conduits for the upward movement of sand and water that emerged as
bedrock until approximately 12 seconds elapsed time (taking the origin sand boils. Other man-made conduits, such as the location of manholes,
of the NAS record as the start of the recording). At that instant, the NAS drop inlets, and risers for pipes, provided similar conduits. Typically, the
site can no longer follow the motion of the bedrock and the two motions pressurized sand and water flowed up along the side of these structures
decouple. It is inferred this decoupling is caused by liquefaction or to the ground surface and migrated out onto the ground surface; Fig. 24
similar of the upper loose soils at 12 seconds, with the upper part of the below illustrates this aspect.
site (including the hangar with the strong motion station) then Two identical two-story concrete-framed office buildings had been
“wobbling” on the softened/liquefied ground largely independent of built within the ITP before 1989. It was recognized that the loose fill
basal acceleration. The ratio of accelerations at the top and bottom parts sands could liquefy, and the buildings were designed with this in mind.
of the soil profile is approximately constant (a factor of 3.75) during the The two structures were supported on spread footings bearing in the
time interval 0–12 seconds of the two records, as observed in Fig. 22, ubiquitous dense sand that caps the site, with footing elevations
where the 3.75 factor is already applied. The decoupling can also be providing a distance of at least twice footing width to the base of the
observed by assessing the change in the frequency content of the ground dense sand. Numerous sand boils were apparent around the ITP build
motion through the use of a Stockwell transform (Stockwell et al. [39]), ings after the earthquake. Ejected sand emerged through a saw cut made
which depicts the time-frequency distribution of a ground motion through the ground floor slab in one building a few days before the
recording. Fig. 23 shows the Stockwell transforms for the NAS and Yerba earthquake, filling the room with a few inches of sand. Earthquake-
Buena records; it can be observed that the large frequencies (i.e., larger induced settlements (estimated from notes made at the time) were in
than 5hz) in the Yerba Buena record are not present in the NAS record the order of 100–150 mm, some of which may have been caused by “loss
after approximately 12 s when the most intense part of the ground of ground” associated with the development of sand boils at the ITP.
motion occurs which may have caused increased excess pore pressures
and a loss in stiffness, which in turn shifts the motion to higher periods 8. Site liquefiability
(lower frequencies). Similar observations were made for ground motions
recorded at liquefied sites by Kramer et al. [40] and Macedo and Bray 8.1. Cyclic stress ratios
[41].
We also focus on the ground motion between recorded times of 8 The shear stresses at a given depth were estimated as τ = agσv rd where
s–16 s, which is the portion of the record where the strongest motion a is the acceleration time history at the surface, σv is the total stress at a
occurred. We estimated that there are about 5 ‘significant’ cycles within given depth, and rd is a depth-dependent factor that modifies the a to
this portion of the record which were subsequently followed by much- account for the flexibility in the dynamic response of a soil mass. We
reduced amplitude motions for a further 10 seconds or so. The num used the rd factors from Boulanger and Idriss [42]. The estimated τ
ber of ‘significant’ loading cycles is about half what would usually be time-history at the middle of layer 5 (~7.5 m depth) is presented in
associated with an earthquake of Loma Prieta’s magnitude, and it may Fig. 25. The cyclic stress ratio (CSR) at a given depth was estimated as
be related to the particularities associated with the fault rupture in the CSR = 0.65 τσmax′ , where σv is the effective vertical stress at a given depth.
′
Fig. 22. Comparison of ground motions at the NAS recording station with the Liquefaction assessment charts/procedures, and strength trends, are
bedrock ground motion recorded at the YBI station. normalized to 15 loading cycles. As there were approximately 5 cycles in
11
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
Fig. 23. Top: Stockwell transform of the acceleration-time history recorded at the top of the NAS station. Bottom: Stockwell transform of the acceleration-time
history recorded at the Yerba Buena station.
Fig. 24. Example of liquefied sand transported to the surface of the building
periphery (1320 Harbor Bay Parkway).
Fig. 25. Estimated cyclic stress ratio history at the middle of Layer 5.
the Loma Prieta record, the sand fill at the ITP will behave ‘stronger’
because of the reduced demand on it. Laboratory data suggest that a 15- the transmitted shear stresses, presents as liquefiable.
to-5 cycle scaling would amount to an apparent strength increase by a
factor of approximately 1.3–1.4 (Jefferies and Been [43]).
8.4. Merritt Sand
The in-situ state parameter of the majority of the loose fill, Layer 2 of
the site characterization (Fig. 3), shows natural variability. Stochastic
In terms of the liquefaction response of the Merritt sand during the
simulations (Popescu et al. [44–47]) indicate that about the loosest 20%
Loma Prieta earthquake, a comparison of the assessed characteristic
of a formation controls its behavior, commonly referred to as the char
state of the Merritt sand (− 0.12 > ψ k > − 0.17) discussed earlier with the
acteristic value from the distribution (and denoted by the subscript ‘k’).
liquefaction triggering criteria shown in Fig. 26, derives in a CRR of
A reasonable judgement is that layer 2 has -0.02 < ψ k < 0.10.
about 0.25, which when compared against the estimated CSR at the top
The assessed demand and in-situ states are shown on a liquefaction
of the Merritt sand, which is in the order of 0.16 using the procedures in
triggering chart, Fig. 26. This chart uses the “Class A” case-histories of
the “cyclic stress ratio” section (after a correction for the number of
Moss [48], updated to a characteristic state parameter (Jefferies and
cycles as discussed in section 8.1), shows the Merritt sand as not lique
Been [43]), which is only valid for F < 1.3. Most, but not all, of the data
fiable. A further factor is that the Merritt sand presents on the soil
for Layer 2 fall within this friction ratio limit of validity (Fig. 12). The
behavior type chart (Fig. 21) as relative compressible, a factor that will
loose hydraulic fill, despite the potential role of the Bay Mud on limiting
also attenuate induced excess pore pressures during cyclic loading. Thus,
12
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
the transmitted stresses, the evidence of sand boils (Fig. 24) suggests
that this was not sufficient to protect the hydraulic sand fill: there was
sand liquefaction at the ITP. This should not come as a surprise as, even
after scaling for the short duration of the Loma Prieta event, the in-situ
state of the fill sands was such that the onset of earthquake-induced
liquefaction was to be expected.
10. Conclusion
13
J. Macedo et al. Soil Dynamics and Earthquake Engineering 158 (2022) 107280
[6] Casagrande A. Characteristics of cohesionless soils affecting the stability of slopes [29] Meehan CL, Duncan JM, Brandon TL, Boulanger RW. An experimental study of the
and earth fills. Harvard University; 1936. dynamic behavior of slickensided surfaces. Report of a study performed by the
[7] Casagrande A. Liquefaction and cyclic deformation of sands, a critical review. Virginia. Tech Center for Geotechnical Practice and Research; 2006.
Buenos Aires: Proceedings of 5th Pan American Conference Soil Mechanic, [30] Campanella RG, Robertson PK, Gillespie D. Seismic cone penetration test. Use of in-
Foundation Engineering; 1975. p. 80–133. situ tests in geotechnical engineering (GSP 6). Reston/VA: ASCE; 1986. p. 116–30.
[8] Taylor DW. Fundamentals of soil mechanics. New York: John Wiley; 1948. [31] Shuttle D, Jefferies M. Determining silt state from CPTu. Geotechn Res 2016;3(3):
[9] Bishop AW. Reply to discussion on “Measurement of shear strength of soils” by A. 90–118. https://doi.org/10.1680/jgere.16.00008. Available at:.
W. Skempton and A.W. Bishop. Géotechnique 1950;2:90–108. [32] Plewes HD, Davies MP, Jefferies MG. CPT based screening procedure for evaluating
[10] Been K, Jefferies MG. A state parameter for sands. Geotechnique 1985;35(2): liquefaction susceptibility. In: Proceedings of the 45th Canadian geotechnical
99–112. Conference; 1992. Canada, Toronto.
[11] Jefferies MG. On the fundamental nature of the state parameter. Géotechnique [33] Reid D. Estimating slope of critical state line from cone penetration test—an
2021. https://doi.org/10.1680/jgeot.20.P.228. update. Can Geotech J 2015;52(1):46–57. https://doi.org/10.1139/cgj-2014-
[12] Morgenstern N, Vick S, Viotti C, Watts B. Fundão tailings dam review panel report 0068.
on the immediate causes of the failure of the Fundão dam. Samarco S.A., Vale S.A; [34] Been K, Jefferies MG. Towards systematic CPT interpretation. In: Proceedings of
2016. the Wroth memorial symposium. London: Thomas Telford; 1992. p. 121–34.
[13] Morgenstern NR, Jefferies M, Zyl DV, Wates J. Independent technical review [35] Mitchell JK, Wentz Jr FJ. Performance of improved ground during the loma Prieta
board. Report on NTSF embankment failure. Ashurst Australia; 2019. Available at: earthquake. University of California Berkeley, Earthquake Engineering Research
https://www.newcrest.com/sites/ default/files/2019-10/190417_Report%20on% Center; 1992. Report No. UCB/EERC-91/12.
20NTSF%20Embankment%20Failure%20at%20Cadia %20for%20Ashurst.pdf. [36] Reid D, Jefferies M. State parameter as a geological principle in tailings. In:
[14] Robertson PK, De Melo L, Williams DJ, Wilson GW. Report of the expert panel on Wilson G Ward, Sego DC, Beier NA, editors. Proceedings of tailings and mine waste
the technical causes of the failure of Feijão dam I. 2019. Available at: http://www. 2017. Edmonton: University of Alberta Geotechnical Center; 2017. p. 305–14.
b1technicalinvestigation.com/. [37] Jefferies MG. Nor-Sand: a simple critical state model for sand. Geotechnique 1993;
[15] Arroyo M, Gens A. Computational analyses of dam I failure at the Corrego de Feijao 43(1):91–103. https://doi.org/10.1680/geot.1993.43.1.91\. Available at:.
mine in Brumadinho – final report. Vale S.A; 2021. [38] Ghafghazi M, Shuttle D. Interpretation of sand state from cone penetration
[16] Stamatopoulos CA. An experimental study of the liquefaction strength of silty sands resistance. Geotechnique 2008;58(8):623–34. https://doi.org/10.1680/
in terms of the state parameter. Soil Dynam Earthq Eng 2010;30(8):662–78. geot.2008.58.8.623.
[17] López-Querol S, Coop MR. Drained cyclic behaviour of loose Dogs Bay sand. [39] Stockwell R, Mansinha L, Lowe RP. Localization of the complex spectrum: the S
Geotechnique 2012;62(4):281–9. transform. IEEE Trans Signal Process 1996;44(4):998–1001. https://doi.org/
[18] Qadimi A, Mohammadi A. Evaluation of state indices in predicting the cyclic and 10.1109/78.492555.
monotonic strength of sands with different fines contents. Soil Dynam Earthq Eng [40] Kramer S, Sideras S, Greenfield M. The timing of liquefaction and its utility in
2014;66:443–58. liquefaction hazard evaluation. Soil Dynam Earthq Eng 2016;91:133–46. https://
[19] Pan K, Yang ZX. Effects of initial static shear on cyclic resistance and pore pressure doi.org/10.1016/j.soildyn.2016.07.025.
generation of saturated sand. Acta Geotechn 2018;13(2):473–87. [41] Macedo J, Bray J. Key trends in assessment of Liquefaction-Induced building
[20] Wei X, Yang J. Cyclic behavior and liquefaction resistance of silty sands with settlement from numerical analyses. Am Soc Civ Eng (ASCE/JGGE) J 2018;144
presence of initial static shear stress. Soil Dynam Earthq Eng 2019;122:274–89. (11). https://doi.org/10.1061/(ASCE)GT.1943-5606.0001951.
[21] Rogers JD, Figuers SH. Engineering geologic site characterization of the greater [42] Boulanger R, Idriss IM. CPT-Based liquefaction triggering procedure. J Geotech
Oakland Alameda area, Alameda and san Francisco Counties, California. Report to Geoenviron Eng 2016;142(2).
the National Science Foundation, report number: Grant No. BCS 9003785. 1992. [43] Jefferies MG, Been K. Soil liquefaction: a critical state approach. second ed. Boca
https://doi.org/10.13140/RG.2.1.4785.6809. Raton, FL, USA: CRC Press; 2015. https://doi.org/10.1201/b19114. Available at:.
[22] Langan Engineering and Environmental Services, Inc. Geotechnical investigation [44] Popescu R, Prevost JH, Deodatis G. Effects of spatial variability on soil liquefaction:
Harbor Bay South Loop. Report to srmErnst Development Partners. Alameda some design recommendations. Geotechnique 1997;475:1019–36.
California; 29 March 2019. [45] Popescu R, Prevost JH, Deodatis G. Spatial variability of soil properties: two case
[23] USGS. U.S. Geological Survey. 1992. https://pubs.er.usgs.gov/publication studies. Geotechnical earthquake engineering and soil dynamics. Geotechnical
/70043734. Special Publication No. 75. Seattle: ASCE; 1998. p. 568–79.
[24] Jefferies M, Been K. Soil variability and characteristic states. Soil liquefaction: a [46] Popescu R, Deodatis G, Prevost JH. Simulation of non-Gaussian homogeneous
critical state approach. CRC Press; 2016. p. 203–24. stochastic vector fields. Probabilist Eng Mech 1998;13(1):1–13.
[25] Jefferies MG, Been K. Soil liquefaction – a critical state approach. London and New [47] Popescu R, Prevost JH, Deodatis G. Characteristic percentile of soil strength for
York: Taylor & Francis Group; 2006, ISBN 0-419-16170-8. dynamic analyses. Geotechnical Earthquake Engineering and Soil Dynamics.
[26] Sitar N, Salgado R. Behavior of the san Francisco Bay Mud from the Marina district Geotechnical Special Publication No. 75. ASCE; 1988. p. 1461–71.
in static and cyclic simple shear. Geotechnical Engineering Report No. UCB/GT/ [48] Moss RES. CPT-based probabilistic assessment of seismic soil liquefaction
93-03. Berkeley, CA: Department of Civil Engineering, University of California; initiation. Ph.D. dissertation. University of California Berkeley; 2003.
April 1993. p. 20. [49] Idriss IM, Boulanger RW. Soil liquefaction during earthquake. Oakland, California,
[27] Rau GA, Sitar N. Post-cyclic response of Holocene Bay Mud from san Francisco. USA: EERI publication, Monograph MNO-12, Earthquake Engineering Research
Geotechn Earthq Eng Soil Dynam 1998;III:246–57. Institute; 2008.
[28] Parks MC. Engineering properties and geologic setting of old Bay clay deposits. [50] Cubrinovski M. Some important considerations in the engineering assessment of
Downtown San Francisco, California: UC Berkeley; 2019. ProQuest ID: Parks_ soil liquefaction. In: Acosta-Martinez H, Lehane BM, editors. Proc., 13th Australia
berkeley_0028E_19138. Merrit ID: ark:/13030/m5jx3mp6. Retrieved from, htt New Zealand Conf. on geomechanics; 2019. p. 19–36.
ps://escholarship.org/uc/item/7zx826gw.
14